Path to this page:
Subject: CVS commit: pkgsrc/databases/unixodbc
From: Ryo ONODERA
Date: 2015-04-26 13:37:08
Message id: 20150426113708.28BD098@cvs.netbsd.org
Log Message:
Update to 2.3.2
* Bump BUILDLINK_API_DEPENDS.
Changelog:
8th-October-2013
2.3.2
* The logging of WStrings was using the incorrect length in some cases
* Pass SQLDescribeCol call to driver when in state 2 (not a cursor spec).
* Pass SQLMoreResults call to driver when in state 2 (not a cursor spec).
Both the last two changes are not as per the original book state table
but allign with the current MS driver manager
* The -e option to isql got lost somewhere. Back in now
* Update install-sh
* SQLCancel assumed that the DM was being built with thread support
* Try and speed up SQLTransact and SQLEndTran operation
* Add missing \ in Postgres7.1 Makefile
* Correct some potential buffer overflows
* Handle SQL_NEED_DATA from a SQLMoreResults
* Get the local charset via nl_langinfo(CODESET) when asking the DM to do \
ASCII-UNICODE conversions
* Handle (and remove) leading spaces from ini entries
* Fixed unicode conversion problems in SQLGetDiagField(W)
* missing terminating null in iusql
* add to the list of errno states that does not cause a create of the ini file
* SQLSetConnactAttr() -> SQLSetConnectAttrW() was passing incorrect \
string length
* Fix double free in SQLGetDiagFieldW
* Fix Unicode/Ansi conversion problem in SQLGetDiagFieldW.c
* Add support for Driver64 in SQLDriverConnectW
* Add missing unicode setting when returning a connection to the pool
* Tidy up leaking iconv handles if connect_part_one fails
* Fix (and avoid) some out of memory problems
* Wrap lt_dlinit and dlerror in the lib mutex
* Add slencheck executable to try and find the sizeof(SQLLEN) from a \
installed driver
* SQL_NO_DATA after SQL_STILL_EXECUTING in SQLExtendedFetch was not setting \
the state correctly
* A little more 3.80 stuff being added
* Added fixes found by coverity
* Added man pages
* Patches to update VMS build
* Change mutex protection around release_env
* Altered strlen to be count of bytes in SQLGetDiagFieldW
* Add check for W function support in do_attr
* Allow SQLDrivers to return attribute length with no supplied buffer
26th-November-2011
2.3.1
* Change type definition of a integer in SQLConnect.c, just to avoid
confusion
* Allow setting the DM overrive values in the connection string to \
SQLDriverConnect
for example
"DRIVER={Easysoft ODBC-SQL \
Server};Server=myserver;UID=user;PWD=pass;DMStmtAttr=SQL_QUERY_TIMEOUT=10;"
* Error and info message order was being inverted by the driver manager
* Fix memory leak in SQLDriverConnect.c (Thanks JM)
* The keyword matching for DRIVER=, DSN= etc was case sensitive. Make it \
insensitive now
* Avoid sprintf NULL pointer problem in SQLGetDiagRec
* Fix typo affecting the pooling of connections, (thanks Chris)
* Fix SunCC _mcount problem
* Attempt to stamp version info on the libs generated. There are aps in use \
linked against other
driver managers that expect VERS_3.52
* Fix potential buffer overrun when using SAFEFILE
* Fix mutex problem in the exit from __SQLGetInfo (thanks Richard)
* Allow getting SQL_DM_VER via SQLGetInfo before connecting to a driver
* Generate unixodbc_conf.h using macros to allow cross compiling
* Fix some libltdl problems
* Fix some naming problems with the cursor lib
* Fix odbcinst problems on systems without pwd.h
* Change lib version to 2 to reflect SQLLEN changes in v2.3
* Fix threading problem (thanks Petr Vandrovec)
* Allow use of lib name in a DRIVER= connection string
* Change default threading protection to 0, most drivers should be thread
safe by now. If the driver is at all thread safe, allow SQLCancel to bipass the
interlock.
* Performance change to handle large numbers of connection and statement \
handles better.
Thanks for the change from the folks at Acision.
* Add -k option to isql to treat the DSN as a connection string and use \
SQLDriverConnect
isql -k "DSN=server;UID=test;PWD=test"
* Couple of the SQLSetConnactAttr values are now SQLULEN instead of SQLUINTEGER
* SQLSetConnectAttr was passing a char length instead of a byte length into the
Driver SQLSetConnectAttr when converting from Ansi->Unicode
* Driver version was not being held when a second connection was made to the \
driver
Files: